For the past several years, people have been saying that AI is out for their jobs. A major group that voiced this were truck drivers that worried about being replaced by autonomous trucks. Then came the factory workers who thought they would be replaced by robots. And then the service workers. Thoughts on AI doing major disruption has led to fear mongering. Not that it’s all wrong. These kinds of things take more time than what’s been imagined. Someday, these particular fears could show to be a real threat.

Generation Gap

Talking about Generative AI in particular: a good chunk of folks are turning to Generative AI for stuff like digging up info (37%), sorting out emails (43%), and whipping up reports (17%). But when it comes to who’s actually tapping into AI, there’s a bit of a generation gap. About 80% of the younger crowd, those between 18 and 24, are all about using AI regularly. On the flip side, nearly two-thirds (63%) of the more seasoned pros, those 55 to 64, haven’t even given AI a whirl.

Freelancer Changes

There’s been a clear change that shows just what an impact Open AI’s GPT has had, though. There’s an indication using data from Upwork job figures that the landscape of freelancer hires has changed due to GPT. The changes are meaningful and point to GPT making a real change here, with video editing/production hires increasing up to 40%, and several other areas, such as accounting, sales, market research, and software-related jobs showing increases. Writing jobs, though, have plummeted over 30%, followed by translation, customer service and social media marketing jobs.

What does this all amount to?

What’s happening now are fears about Generative AI. It’s so resourceful. The fear is that it will replace large numbers of workers in different fields. And those fears are right, but also wrong. Here’s what we mean.

Entertainment Industry

• There will be certain areas of work, certain business verticals, that will be affected much more than others by Generative AI, so much so that many jobs will be eliminated as a result. For other verticals, we believe that Generative AI will be used to augment existing jobs, though job loss will occur there as well, but not as much. We talked about the entertainment industry last week and what text-to-video means there. So, the entertainment industry will be one of the heaviest hit. And it’s not all about video, it’s scriptwriting, replacement/digital duplicates of actors and actresses, people who do CGI, storyboarding, gaming, etc. Basically, everything except the deal-making. And then there’s the big one—customer service, where chatbots and personalized assistants can literally replace large groups of people. Companies that have used Generative AI include Warner Bros., Disney, YouTube, Netflix, and Amazon Prime Video. Generative AI companies in entertainment include Runway, Flawless, Synthesia, Metaphysic, and Wonder Dynamics.

Legal Industry

• Another area is legal, where paralegals could be replaced in large part by generative AI being able to scan immense amounts of contracts and can also create legal documents, as well as simulate legal outcomes to aid in case preparation. We know of several legal firms that are testing Harvey that builds custom LLMs for law firms. In particular, the law firms are one of the most paranoid about Generative AI, because they fear that someday lawyers might be replaced by this technology. So, they are one of the most reticent in terms of adopting Generative AI. And yet, they are testing the use of it, because there are many cost benefits to them. Besides Harvey, Generative AI companies in the legal industry include Eve, Casetext, and DoNotPay. In terms of which law firms are using Generative AI, that is currently under wraps, as law firms are conservative and do not want to be known as using AI.

Finance industry

• That’s what it comes down to for businesses and Generative AI—saving money by cutting out costs, as well as the time it takes to do tasks. The finance sector sees AI impacting roles such as financial analysis, where algorithms can perform data analysis and predictive modeling. Jobs in areas like claims processing and underwriting are also at risk because of AI's ability to automate these processes. Companies that have used Generative AI include JP Morgan Chase, Goldman Sachs, Morgan Stanley, Bloomberg, and NatWest Group. Generative AI companies in the finance industry include Finpilot, Finance GPT and Midday.

Real Estate and Construction

• Other verticals affected are real estate and construction, where in the future, architectural models could be produced on the fly and those who do market analysis could be replaced. For instance, some companies are already doing automated structural engineering like Hedral. Here are seven AI companies working on real estate: https://twitter.com/i/lists/1705718284488986722. Companies that have used Generative AI include Zillow, Redfin, StreetEasy, Suffolk and Black and McDonald.

Technology and Electronics

• It’s hard to overstate how Generative AI is affecting the software world. Whole code could be created as well as augmented and debugged by Generative AI. We have personally used GPT-4 as a test case for this and for writing whole code it is not the bomb. It works best currently for debugging and even then, several times were needed to get some code properly debugged. And yet, there are many coders that are using Generative AI as a shortcut because overall, it could save time. In the future, we see Generative AI being able to proficiently create whole code. Besides GPT-4, another Generative AI tool for code completion is GitHub Copilot. Companies that have used Generative AI code completion include JPMorgan Chase, Goldman Sachs and Morgan Stanley.

• What does this mean? Whole groups of staff engineers could potentially be replaced. Perhaps they could be used for checking the code created by Generative AI. But what’s for sure is that there will be replacements due to Generative AI. In electronics, components like chips and circuit boards could potentially be designed by Generative AI and people who do this will be at a risk. However, there will be new kinds of jobs that will be created—more on this coming up. Companies that have used Generative AI for chip design include Synopsys and Cadence Design Systems.

Marketing and Advertising

Other verticals include marketing and advertising, human resources, education, and news media. 

• For marketing and advertising, the effects of Generative AI are really clear. Why have people labor over creating visuals for campaigns when Generative AI can do that? Not only visuals but copy for those campaigns. So, large groups of people that have been creating these historically without Generative AI we see as being easily replaced. Sure, there will be people needed to oversee what Generative AI has produced. Here are 200 AI Companies helping with AI writing or marketing: https://twitter.com/i/lists/1705715250895667539?s=20. Companies that have used Generative AI include WPP, Unilever, Nestlé and Mondelez.

Human Resources

• For human resources, an area which has already been heavily impacted by digitization, Generative AI will put the final nail in the coffin for people who have been judging individual resumes one-by-one. Companies that have used Generative AI for human resources include Amazon, PayPal, Google and Uber. Generative AI companies include Instahyre and paidleave.

   

Education

• A large part of teaching assistant jobs could be taken over by Generative AI—in creating lesson plans and tests, as well as the grading of tests, including essays. Here are 93 AI companies working on education: https://twitter.com/i/lists/1705702737835643273?s=20. Companies that have used Generative AI include Pearson, McGraw Hill, Cengage and Knewton.

   

News Media

• What about the news industry and reporting? The use of Generative AI was quickly retracted by news organizations because of the errors and hallucinations done by Generative AI. These made it into news stories that were published. What had not been foreseen is that there needed to be intense scrutiny of these Generative AI-created news stories by people to make sure that they were accurate and error-free. That is the situation now, but we feel that in the future Generative AI will get much better in terms of its hallucination problem so that news organizations will start using it in a large way to produce news stories. There should always be some kind of overview of these generated news stories, but we might be asking too much in a much more automated future, though Business Insider, Forbes and Trusted Media Brands have said that more caution needs to be taken. Besides Forbes, Insider, and Trusted Media Brands, The New York Times has used Generative AI. A Generative AI company in news media that is being used is Nota.

Measured Potential Impact of Generative AI

The chart below shows just what the potential of generative AI could have on automation in global employment activities and it’s really stunning. The study was done by McKinsey & Co. 

What Do We Really Think About it?

So, these are the areas of employment most affected by Generative AI. What is going to come of all of this?

• It will do a great deal of good in general. Instead of taking away more jobs, it will introduce new kinds of jobs and augment the ones that already exist. Generative AI is a tool to help people do their jobs better.

In the art world, there has been a fear that digitization will replace artistic creativity over the past several decades and Generative AI has now increased this ten-fold. But, in talking with many artists who are using Generative AI, what is clear is that Generative AI is a real tool for them. It’s a new type of paint brush and oil for painting. The medium is different, just like photography is different from drawing, but the art that people create themselves is there.

This is just an example of how we see that jobs will be augmented by the use of Generative AI. Even in marketing and advertising, the most talented in charge of accounts will be able to use Generative AI without being replaced.

• What we are saying is, if you don’t learn how to use Generative AI, your job might be in jeopardy. But, this is similar to learning how to use a computer over a typewriter, or better yet, replacing the horse with the automobile. How could you not do this? You have to.

In terms of new kinds of jobs being created than there were before due to Generative AI, there’s a whole slew of people that will be hired to oversee how and what Generative AI is doing. And this is in all the verticals mentioned. How many years until this is no longer needed? We don’t see this oversight as ever stopping, so the new types of jobs created will be permanent and long-going.

Looking Forward

Looking further in the future, when robots are mainstays in homes—these robots will be using Generative AI or something like it to do their jobs and interface with people. They will need constant oversight to make sure they are performing correctly and that the people who are using the robots are safe.

That is a futuristic view (we see this happening in about 15 years), but coming back to the present—how comfortable do we think it will be for people to use Generative AI in their jobs? Currently, it is a little difficult. The user interface is not easy. You need to know how to prompt well and even then, it would be a wildcard in terms of what you produce with Generative AI. Though, there is a growing amount of what is called “fine-tuning”—the creation of smaller, narrow LLMs created for certain verticals, such as the legal industry. We mentioned Harvey here already. It is a narrow LLM that is finetuned to a particular law firm’s database and other wanted documentation, so that a person using it will get results that are relevant and nothing extraneous.

This fine-tuning is being done across many verticals, including healthcare. There is a particular company called Hippocratic AI that has fine-tuned their LLM across nurse paperwork so that a level of care could be made by an LLM rather than a nurse through a chatbot. There are companies who are fine-tuning across their radiological databases, so that initial diagnoses could be made using a doctor’s discretion.

And that is the word – discretion. Discretion and oversight are key.

Why Do it?

• So, how easy is it to learn and use Generative AI right now? In a few words—not easy. Yet, we are saying that you should start to learn and use it so as not to be left behind.

Here are some more examples of how using Generative AI can augment someone’s job.

Even in the entertainment industry, where there is so much at stake, using Generative AI could be immensely helpful, not hurtful, to existing employees. Let’s say that you are a writer for an episodic TV show and you are looking for an idea for a storyline for the next episode. You could use Generative AI to do that and then write it up yourself. This could also be used for figuring out how dialogue should go. Basically, it’s to spur creativity, not replace it.

Now, let’s say you are a director and you want to storyboard the scenes you see in your head. You could use Generative AI—text-to-image—to visualize that and show it to others.

Generative AI could also be used in gaming to aid in storyboarding and dialogue, as well as coding, other visuals as a part of character development, and audio. This augments what a person is already doing and is helpful.

We mentioned audio. Musicians could use Generative AI to test out a new tune they are thinking of recording, and also duplicate their voice for their own creative use.

Now, another way this could go for the entertainment industry is that actors, actresses and musicians could license out their digital doubles that are created by Generative AI. This is already starting to be done, with the artists’ agency, CAA, spearheading it. So, new business models will also be created with the use of Generative AI.

And along those lines, let’s say that when text-to-video allows for longer than 60-second creations, there will be creators that will need to showcase their works—so a new kind of distribution. Will YouTube or Netflix be the company which will come out as the winner here or will there be new companies that are created to take advantage of this new opportunity? 

• So, to summarize what we have said here. So many people are afraid of what Generative AI will do to their jobs. And in some verticals, there is definitely a chance of Generative AI taking away a job or a large chunk of a job. But, we see Generative AI as a glass half-full for existing employees. We think it is important to start to learn Generative AI now, even though the user interface is not easy. So, when it does get easier, you already have a leg up on understanding and using Generative AI in your job.

• Generative AI is a tool to be used, not a replacement device. That’s how it is with technology in general. Technology is a tool—better find a way to use it!

Just Three Things

According to Scoble and Cronin, the top three relevant happenings last week

Google Gemini AI

Google put a pause on Gemini, its Generative AI tech, from producing images of people after its algorithms misrepresented historical figures as being people of color. U.S. Founding fathers and Nazis were generated as non-white people, and in some cases, as women. Google has not directly addressed how these anomalies could happen. We understand how difficult it is to make sure that Generative AI systems are not biased; this is an example of trying to fix something and causing other issues. We feel that ultimately generative AI companies will get it right. NBC News

First Neuralink Human Trial

On an X Space, Elon Musk announced that the first human patient that was given a Neuralink implant can control a mouse’s movement just by using thought. There has been no scientific documentation produced yet, but it will be a big win for the company, if verified, though other companies are currently ahead of Neuralink. We feel that with Musk’s perseverance and success rate, Neuralink will come out as the winner in their field. CNN

Groq

Groq, an AI chip company that produces Language Processing Units (LPUs), in a surprise move went live with a demo posted on X that went viral and wowed everyone. Their claim is that they could deliver “the world’s fastest large language models.” According to a third-party evaluation, Groq achieves a rate of 247 tokens per second, vastly outpacing Microsoft's 18 tokens per second. This also implies that ChatGPT could operate over 13 times faster on Groq's chips. The game is on for the chip race to win AI dominance! Gizmodo